WelcomeWelcome | FAQFAQ | DownloadsDownloads | WikiWiki

Author Topic: Odd kernel behavior or motherboard?  (Read 2147 times)

Offline CentralWare

  • Retired Admins
  • Hero Member
  • *****
  • Posts: 776
Odd kernel behavior or motherboard?
« on: December 30, 2020, 10:55:06 AM »

TCL v11 x64


I have a Gigabyte motherboard, 2016 AM BIOS (no vendor updates I'm aware of beyond what's flashed) and when booting TCL v11 CorePure64, Ubuntu Mate and CentOS-8 (Boot) via USB2 they all have the same issue in common: No eth0 and no USB3.x (which I'm guessing ethernet is fed from the USB3 bridge.)  When I installed Win10 on the machine, these issues didn't exist.


1. After researching this particular motherboard I came to find a few tweaks to BIOS settings that were supposedly required (reverted back to Windows 8 mode with CSM, disabled EHCI pass-through and updated the kernel line to include both amd_iommu=on and iommu=pt (any idea if adding intel_iommu would cause issues - for a more "portable" USB stick?)


2. After all of the above changes, I still have no ethernet/usb3.  I then downloaded all of the firmware-* and *-KERNEL.tcz files I could think of which were network/usb related; i2c-KERNEL seemed to have made it possible to "see" eth0 prior to the iommu entries on the command-line, but the hardware kept looping/resetting.  When the entries were added, eth0 vanished.


3. Finally, after downloading/installing all modules possible, eth0 came to life - and worked.  Process of illumination shows a very strange dependency though which I figured I'd ask to see if anyone had a clue.  I have everything removed from onboot.lst except graphics-KERNEL.tcz (which complains/panics if I don't have Raedon, too) - which the machine has two PCIe x16 AMD cards installed which shouldn't have anything to do with the onboard NIC/USB3.x PCIe channel - but without the graphics module, nada.


Any clues how they're related?  (Note: I haven't yet tested earlier kernels to see if there's a starting point with this issue or whether or not it seems to be motherboard specific.)


Thanks gang!


Offline Rich

  • Administrator
  • Hero Member
  • *****
  • Posts: 11703
Re: Odd kernel behavior or motherboard?
« Reply #1 on: December 30, 2020, 11:14:03 AM »
Hi centralware
...  which the machine has two PCIe x16 AMD cards installed which shouldn't have anything to do with the onboard NIC/USB3.x PCIe channel - but without the graphics module, nada. ...
One way to find out. Pull one of the PCIe x16 AMD cards and see if anything changes.

Offline CentralWare

  • Retired Admins
  • Hero Member
  • *****
  • Posts: 776
Re: Odd kernel behavior or motherboard?
« Reply #2 on: December 30, 2020, 01:31:26 PM »
No change.  (The board doesn't have an onboard GPU; not sure if that matters.)
https://www.gigabyte.com/Motherboard/GA-970-Gaming-rev-10
I'm going to up/downgrade BIOS to see if it has any effect...


Offline CentralWare

  • Retired Admins
  • Hero Member
  • *****
  • Posts: 776
Re: Odd kernel behavior or motherboard?
« Reply #3 on: December 30, 2020, 03:00:12 PM »
Up/downgrade didn't do anything (other than scare the #### out of me - booted into 11.x/x86 and the screen graphics were waaay out of sorts - reminds me of the ASCII days ;) turns out bios flash back to 2015 sent all settings back to factory default.)


After digging a little, it seems as though the PCIe channel for ethernet/USB3.0/USB3.1 is accessed by i2c...  and the Radeon firmware happens to have i2c as a dependency.  Something in the graphics module has the missing kernel driver piece needed for them all to play nice together and come alive.  Cannot fathom why!?